rlucene 0.1.0-alpha.3

A Rust port of Apache Lucene, a full-text search engine library
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
use crate::core::codecs::knn_vectors_reader::KnnVectorsReader;
use crate::core::index::field_info::FieldInfo;
use crate::core::index::knn_vector_values::{
  DocIndexIterator, DocIndexIteratorEnum2, KnnVectorValues,
};
use crate::core::index::merge_state::DocMap;
use crate::core::index::vector_encoding::VectorEncoding;
use crate::core::search::doc_id_set_iterator::DocIdSetIterator;
use crate::core::search::doc_id_set_iterator::NO_MORE_DOCS;
use crate::core::util::bit_set::BitSet;
use crate::core::util::bits::Bits;
use crate::core::util::concurrent_hnsw_merger::ConcurrentHnswMerger;
use crate::core::util::error::lucene_error::{LuceneError, Result};
use crate::core::util::fixed_bit_set::FixedBitSet;
use crate::core::util::hnsw::hnsw_builder::HnswBuilder;
use crate::core::util::hnsw::hnsw_graph_builder::{create_with_graph_size, rand_seed};
use crate::core::util::hnsw::hnsw_graph_merger::HnswGraphMerger;
use crate::core::util::hnsw::initialized_hnsw_graph_builder::from_graph;
use crate::core::util::hnsw::on_heap_hnsw_graph::OnHeapHnswGraph;
use crate::core::util::hnsw::random_vector_scorer_supplier::RandomVectorScorerSupplier;
use crate::core::util::info_stream::InfoStreamMT;
use std::collections::HashMap;
use std::sync::Arc;

#[derive(Default)]
pub(crate) enum HnswGraphMergerHook {
  #[default]
  Default,
  Concurrent(ConcurrentHnswMerger),
}

pub(crate) struct IncrementalHnswGraphMergerDefaults;

pub(crate) trait IncrementalHnswGraphMergerBase<S>
where
  S: RandomVectorScorerSupplier,
{
  #[allow(clippy::too_many_arguments)]
  fn create_builder<KV, R, D>(
    &self,
    field_info: &FieldInfo,
    scorer_supplier: S,
    m: usize,
    beam_width: usize,
    init_reader: Option<usize>,
    init_doc_map: Option<usize>,
    init_graph_size: usize,
    merged_vector_values: &mut KV,
    max_ord: i32,
    readers: &[Option<R>],
    doc_maps: &[D],
    info_stream: InfoStreamMT,
  ) -> Result<OnHeapHnswGraph>
  where
    KV: KnnVectorValues,
    R: KnnVectorsReader,
    D: DocMap,
  {
    IncrementalHnswGraphMergerDefaults::create_builder(
      field_info,
      scorer_supplier,
      m,
      beam_width,
      init_reader,
      init_doc_map,
      init_graph_size,
      merged_vector_values,
      max_ord,
      readers,
      doc_maps,
      info_stream,
    )
  }
}

impl<S> IncrementalHnswGraphMergerBase<S> for HnswGraphMergerHook
where
  S: RandomVectorScorerSupplier,
{
  fn create_builder<KV, R, D>(
    &self,
    field_info: &FieldInfo,
    scorer_supplier: S,
    m: usize,
    beam_width: usize,
    init_reader: Option<usize>,
    init_doc_map: Option<usize>,
    init_graph_size: usize,
    merged_vector_values: &mut KV,
    max_ord: i32,
    readers: &[Option<R>],
    doc_maps: &[D],
    info_stream: InfoStreamMT,
  ) -> Result<OnHeapHnswGraph>
  where
    KV: KnnVectorValues,
    R: KnnVectorsReader,
    D: DocMap,
  {
    match self {
      Self::Default => IncrementalHnswGraphMergerDefaults::create_builder(
        field_info,
        scorer_supplier,
        m,
        beam_width,
        init_reader,
        init_doc_map,
        init_graph_size,
        merged_vector_values,
        max_ord,
        readers,
        doc_maps,
        info_stream,
      ),
      Self::Concurrent(hook) => hook.create_builder(
        field_info,
        scorer_supplier,
        m,
        beam_width,
        init_reader,
        init_doc_map,
        init_graph_size,
        merged_vector_values,
        max_ord,
        readers,
        doc_maps,
        info_stream,
      ),
    }
  }
}

/// This selects the biggest Hnsw graph from the provided merge state and initializes a new
/// HnswGraphBuilder with that graph as a starting point.
pub struct IncrementalHnswGraphMerger<S> {
  field_info: Arc<FieldInfo>,
  scorer_supplier: Option<S>,
  m: usize,
  beam_width: usize,
  init_reader: Option<usize>,
  init_doc_map: Option<usize>,
  init_graph_size: usize,
  hook: HnswGraphMergerHook,
}

impl<S> IncrementalHnswGraphMerger<S> {
  pub fn new(field_info: Arc<FieldInfo>, scorer_supplier: S, m: usize, beam_width: usize) -> Self {
    Self {
      field_info,
      scorer_supplier: Some(scorer_supplier),
      m,
      beam_width,
      init_reader: None,
      init_doc_map: None,
      init_graph_size: 0,
      hook: HnswGraphMergerHook::Default,
    }
  }
}

impl<S> IncrementalHnswGraphMerger<S> {
  pub(crate) fn new_with_hook(
    field_info: Arc<FieldInfo>,
    scorer_supplier: S,
    m: usize,
    beam_width: usize,
    hook: HnswGraphMergerHook,
  ) -> Self {
    Self {
      field_info,
      scorer_supplier: Some(scorer_supplier),
      m,
      beam_width,
      init_reader: None,
      init_doc_map: None,
      init_graph_size: 0,
      hook,
    }
  }
}

impl<S> IncrementalHnswGraphMerger<S>
where
  S: RandomVectorScorerSupplier,
{
  /// Builds a new HnswGraphBuilder using the biggest graph from the merge state as a starting point.
  /// If no valid readers were added to the merge state, a new graph is created.
  ///
  /// # Arguments
  ///
  /// * `merged_vector_values` - vector values in the merged segment
  /// * `max_ord` - max num of vectors that will be merged into the graph
  ///
  /// # Returns
  ///
  /// HnswGraphBuilder
  ///
  /// # Errors
  ///
  /// Returns an error if reading from the merge state fails.
  fn create_builder<KV, R, D>(
    &mut self,
    merged_vector_values: &mut KV,
    max_ord: i32,
    readers: &[Option<R>],
    doc_maps: &[D],
    info_stream: InfoStreamMT,
  ) -> Result<OnHeapHnswGraph>
  where
    KV: KnnVectorValues,
    R: KnnVectorsReader,
    D: DocMap,
  {
    let scorer_supplier = self
      .scorer_supplier
      .take()
      .ok_or_else(|| LuceneError::illegal_state("scorer supplier has already been consumed"))?;
    self.hook.create_builder(
      self.field_info.as_ref(),
      scorer_supplier,
      self.m,
      self.beam_width,
      self.init_reader,
      self.init_doc_map,
      self.init_graph_size,
      merged_vector_values,
      max_ord,
      readers,
      doc_maps,
      info_stream,
    )
  }
}

impl IncrementalHnswGraphMergerDefaults {
  #[allow(clippy::too_many_arguments)]
  fn create_builder<KV, R, D, T>(
    field_info: &FieldInfo,
    scorer_supplier: T,
    m: usize,
    beam_width: usize,
    init_reader: Option<usize>,
    init_doc_map: Option<usize>,
    init_graph_size: usize,
    merged_vector_values: &mut KV,
    max_ord: i32,
    readers: &[Option<R>],
    doc_maps: &[D],
    info_stream: InfoStreamMT,
  ) -> Result<OnHeapHnswGraph>
  where
    KV: KnnVectorValues,
    R: KnnVectorsReader,
    D: DocMap,
    T: RandomVectorScorerSupplier,
  {
    let max_ord_usize = max_ord as usize;
    match init_reader {
      Some(init_reader_idx) => {
        let init_reader = readers[init_reader_idx].as_ref().ok_or_else(|| {
          LuceneError::illegal_state(format!(
            "Reader at index {init_reader_idx} is not available"
          ))
        })?;
        let mut initializer_graph = init_reader.get_graph(field_info.name.as_str())?;
        let mut initialized_nodes = FixedBitSet::new(max_ord_usize);
        let init_doc_map_idx = init_doc_map
          .ok_or_else(|| LuceneError::illegal_state("initializer reader has no document map"))?;
        let doc_map = doc_maps.get(init_doc_map_idx).ok_or_else(|| {
          LuceneError::illegal_state(format!(
            "DocMap at index {init_doc_map_idx} is not available"
          ))
        })?;
        let old_to_new_ordinal_map = get_new_ord_mapping(
          field_info,
          init_graph_size,
          merged_vector_values,
          &mut initialized_nodes,
          init_reader,
          doc_map,
        )?;
        let mut builder = from_graph(
          scorer_supplier,
          m,
          beam_width,
          rand_seed(),
          &mut initializer_graph,
          &old_to_new_ordinal_map,
          initialized_nodes,
          max_ord,
        )?;
        builder.set_info_stream(info_stream);
        Ok(std::mem::replace(
          builder.build(max_ord_usize)?,
          OnHeapHnswGraph::new(m, 0),
        ))
      },
      None => {
        let mut builder =
          create_with_graph_size(scorer_supplier, m, beam_width, rand_seed(), max_ord)?;
        builder.set_info_stream(info_stream);
        Ok(std::mem::replace(
          builder.build(max_ord_usize)?,
          OnHeapHnswGraph::new(m, 0),
        ))
      },
    }
  }
}

/// Creates a new mapping from old ordinals to new ordinals and returns the total number of vectors
/// in the newly merged segment.
#[allow(clippy::too_many_arguments)]
pub(crate) fn get_new_ord_mapping<KV, R, D>(
  field_info: &FieldInfo,
  init_graph_size: usize,
  merged_vector_values: &mut KV,
  initialized_nodes: &mut FixedBitSet,
  reader: &R,
  init_doc_map: &D,
) -> Result<Vec<usize>>
where
  KV: KnnVectorValues,
  R: KnnVectorsReader,
  D: DocMap,
{
  let mut initializer_iterator = match field_info.get_vector_encoding() {
    VectorEncoding::BYTE(_) => DocIndexIteratorEnum2::A(
      reader
        .get_byte_vector_values(&field_info.name)?
        .iterator()?,
    ),
    VectorEncoding::FLOAT32(_) => DocIndexIteratorEnum2::B(
      reader
        .get_float_vector_values(&field_info.name)?
        .iterator()?,
    ),
  };

  let mut new_id_to_old_ordinal = HashMap::with_capacity(init_graph_size);
  let mut max_new_doc_id = -1;
  let mut doc_id = initializer_iterator.next_doc()?;
  while doc_id != NO_MORE_DOCS {
    let new_id = init_doc_map.get(doc_id)?;
    max_new_doc_id = max_new_doc_id.max(new_id);
    new_id_to_old_ordinal.insert(new_id, initializer_iterator.index()? as usize);
    doc_id = initializer_iterator.next_doc()?;
  }

  if max_new_doc_id == -1 {
    return Ok(Vec::new());
  }

  let mut old_to_new_ordinal_map = vec![0; init_graph_size];
  let mut merged_vector_iterator = merged_vector_values.iterator()?;
  let mut new_doc_id = merged_vector_iterator.next_doc()?;
  while new_doc_id != NO_MORE_DOCS && new_doc_id <= max_new_doc_id {
    if let Some(&old_ordinal) = new_id_to_old_ordinal.get(&new_doc_id) {
      let new_ord = merged_vector_iterator.index()? as usize;
      initialized_nodes.set(new_ord)?;
      old_to_new_ordinal_map[old_ordinal] = new_ord;
    }
    new_doc_id = merged_vector_iterator.next_doc()?;
  }

  Ok(old_to_new_ordinal_map)
}

impl<S> HnswGraphMerger for IncrementalHnswGraphMerger<S>
where
  S: RandomVectorScorerSupplier,
{
  /// Adds a reader to the graph merger if it meets the following criteria: 1. Does not contain any
  /// deleted docs 2. Is a HnswGraphProvider/PerFieldKnnVectorReader 3. Has the most docs of any
  /// previous reader that met the above criteria
  fn add_reader<R, B>(
    &mut self,
    reader_index: usize,
    reader: &R,
    doc_map_idx: usize,
    live_docs: Option<&B>,
  ) -> Result<()>
  where
    R: KnnVectorsReader,
    B: Bits,
  {
    if !reader.is_hnsw_graph_provider(&self.field_info.name) || !no_deletes(live_docs)? {
      return Ok(());
    }

    let candidate_vector_count = match self.field_info.get_vector_encoding() {
      VectorEncoding::BYTE(_) => reader.get_byte_vector_values(&self.field_info.name)?.size(),
      VectorEncoding::FLOAT32(_) => reader
        .get_float_vector_values(&self.field_info.name)?
        .size(),
    };

    if candidate_vector_count <= self.init_graph_size {
      return Ok(());
    }
    self.init_reader = Some(reader_index);
    self.init_doc_map = Some(doc_map_idx);
    self.init_graph_size = candidate_vector_count;
    Ok(())
  }

  fn merge<KV, R, D>(
    &mut self,
    mut merged_vector_values: KV,
    info_stream: InfoStreamMT,
    max_ord: i32,
    readers: &[Option<R>],
    doc_map: &[D],
  ) -> Result<OnHeapHnswGraph>
  where
    KV: KnnVectorValues,
    R: KnnVectorsReader,
    D: DocMap,
  {
    self.create_builder(
      &mut merged_vector_values,
      max_ord,
      readers,
      doc_map,
      info_stream,
    )
  }
}

fn no_deletes<B>(live_docs: Option<&B>) -> Result<bool>
where
  B: Bits,
{
  let Some(live_docs) = live_docs else {
    return Ok(true);
  };

  for i in 0..live_docs.length() {
    if !live_docs.get(i)? {
      return Ok(false);
    }
  }
  Ok(true)
}